Copeptin as a marker of outcome after cardiac arrest: a sub-study of the TTM trial

This is a predefined biomarker sub-study of the target temperature management (TTM) 33 °C versus 36 °C after out-of-hospital cardiac arrest trial [17], randomizing 950 unconscious OHCA patients to an intervention of 33 °C or 36 °C, indicating no significant benefit of a target temperature of 33 °C compared to 36 °C [17]. The trial protocol was approved by ethical committees in each participating country, and informed consent was waived or obtained from all participants or relatives according to national legislation, in line with the Helsinki declaration [18].

Unconscious adult patients were included in the TTM trial within 4 h of return of stable spontaneous circulation (ROSC) after OHCA of a presumed cardiac cause. Previous medical history was recorded without no strict definitions. Patient hemodynamics at screening and before randomization were classified in one of three categories: 1, no shock, systolic blood pressure (SBP) > 90 mmHg; 2, moderate shock, SBP < 90 mmHg for > 30 min, or the need of supportive measures (fluid loading, vasopressor, and/or inotropic support) to maintain SBP > 90 mmHg and/or end-organ hypoperfusion (cool extremities, urine output < 30 ml/h); and 3, severe shock, SBP < 80 mmHg in spite of supportive measures (medical and mechanical) that could not be reversed within the inclusion window. Patients in persistent severe shock were not eligible. Of the 36 study sites, seven sites did not participate in the biobank sub-study due to legal or logistical reasons. This analysis included all patients alive at 24 h with at least one recorded copeptin level during the sampling period.

Blood serum samples were collected at 24, 48, and 72 h after enrolment in the study and were processed at the study sites, aliquoted and frozen to − 80 °C before shipment to the Integrated Biobank of Luxemburg. Centralized batch analysis of copeptin was done using immune fluorescence with the Brahms Kryptor Compact Plus system (Thermo Fisher Scientific Brahms, Germany), with range of 0.7–500 pmol/L (auto dilution with upper range of 2000 pmol/L) and functioning assay sensitivity of 1.08 pmol/L. All serum samples with interference detected by the Kryptor system were discarded. Analyses of copeptin were made at least 6 months after trial completion.

The primary outcome was incidence of death until 30 days after enrolment in the TTM trial. The secondary endpoints were circulatory cause of death within 30 days of cardiac arrest, as estimated from clinical data, and the extended cardiovascular sequential organ failure assessment score (eCvSOFA), a nine level sub-score of the SOFA cardiovascular component [19]. In addition, we devised a binary composite outcome of cardiovascular deterioration (cardiovascular deterioration composite (CvDC)) considered positive if the patient had an eCvSOFA ≥ 5 or died from circulatory cause at within ± 12 h of copeptin sample or if eCvSOFA score increased more than two points within the previous 24 h.

Descriptive statistics were used to summarize the study population. Continuous data are presented as median with interquartile range. Differences in baseline variables were assessed using the Mann-Whitney or the χ² test, as appropriate. Analyses were primarily performed on 24-h samples of copeptin, with the secondary analyses performed on serial samples (48, 72-h samples). No formal adjustment of significance levels due to multitesting was performed. Due to missingness in the dataset, analyses were performed on a pooled dataset for the explanatory models based on 20 imputations by chained equations, using predictive mean matching for continuous variables and logistic regression for categorical data. Because of skewed distribution of copeptin data, log2-transformed copeptin levels were used in Cox and logistic regression models. All models were adjusted for TTM at 33 °C (yes/no) and early clinical predictors of outcome according to the TTM-score [20]: age (years), cardiac arrest at home (yes/no), no-flow time (minutes from cardiac arrest until start of chest compressions or ROSC, whichever was first), low-flow time (minutes with chest compressions), shockable rhythm (yes/no), use of adrenaline during CPR (yes/no), absence of corneal and pupillary reflexes on admission (yes/no), Glasgow coma score motor component (mGCS) on admission > 1 or sedated (yes/no), admission pH, and admission PaCO₂ < 4.5 kPa (yes/no). The log-rank test was used to test for difference in 30-day survival according to copeptin levels stratified as above or below median at the specified time point in the biobank population and visualized in Kaplan-Meier graphs. Adjusted hazard ratio for death was estimated using the Cox regression of proportional hazards model, with follow-up time censored at 30 days. Multivariate logistic regression was performed for circulatory cause of death. Correlation between copeptin and eCvSOFA-score was assessed using Spearman’s rank order correlation. Multivariate logistic regression was performed for CvDC.

A sensitivity analysis was performed comparing the results of a complete cases adjusted Cox regression model for the primary endpoint at 24, 48, and 72 h with that of the imputed model.

Statistical analysis was performed using software R (v 3.4.0) and RStudio (v 1.0.143).

Differences in copeptin levels between interventions were highest at 72 h, 33.00 [15.76–63.62] for 33 °C and 26.46 [11.37–59.28] pmol/l, for 36 °C, p = 0.049 (Additional file 2). Copeptin levels increased more between 24 h and 48 h at an intervention of 33 °C, 6.11 [− 4.25–36.96] compared to 36 °C 1.91 [− 9.51–27.38] pmol/l, (p = 0.026), while no significant differences were seen between 48 and 72 h. Since differences between intervention arms were minor and our primary aim was to test associations between copeptin and outcome, all further analyses were performed on pooled samples.

Two hundred seventy-three (39.6%) patients died within 30 days. Copeptin levels were significantly lower in survivors than in patients dead by day 30, at 24 h 17.19 [9.89–36.20] versus 51.14 [19.60–99.51] pmol/l, at 48 h 31.84 [15.44–59.49] versus 57.11 [24.32–102.97] pmol/l, and at 72 h 22.19 [9.89–48.26] versus 47.37 [22.90–101.78] pmol/l, all p values < 0.001 (Fig. 1). Crude 30-day survival was associated with copeptin stratified as above or below median at 24 h, p < 0.001 (Fig. 2), at 48 h, p = 0.001, and 72 h p < 0.001 (Additional file 3). The incidence of death was independently associated with log2-transformed copeptin, hazard ratio (HR) 1.17 [1.06–1.28], p = 0.001(Fig. 3); for samples at 24 h, significance was lost at 48 h, p = 0.471, while 72-h samples were trending towards significance, HR 1.11 [1.00–1.23], p = 0.054.

In the adjusted model, log2-transformed copeptin was independently associated with circulatory cause of death, at 24 h, with an odds ratio (OR) of 1.03 [1.01–1.04], p = 0.001 (Additional file 4). The Spearman rank order correlation between eCvSOFA and copeptin, rho, was 0.19 [0.12–0.27] at 24 h, 0.22 [0.14–0.29] at 48 h and 0.26 [0.18–0.34] at 72 h, p < 0.001, for all time points. Log2-transformed copeptin was independently associated with cardiovascular deterioration with OR 1.05 [1.02–1.08], p < 0.001 at 24 h (Fig. 4); OR 1.03 [1.00–1.05], p = 0.021 at 48 h; and OR 1.03 [1.01–1.06], p = 0.016 at 72 h.

In the sensitivity analysis, results from the Cox regression models on imputed datasets and observed data were similar (data not shown).

This is a large predefined sub-study of the TTM trial investigating copeptin as a marker of severity of the post cardiac arrest syndrome. The TTM trial had strict rules for withdrawal of life-sustaining therapy, and all clinical data were prospectively collected. Analyses were made in a single laboratory, limiting the risk of inter-laboratory assay variability, and copeptin values were analyzed after trial completion, eliminating the risk of treatment bias. Due to the limitations of an observational study, we can however only assess associations with outcome. It cannot be ruled out that missing data may have affected our results. The collection of blood samples for the biobank of the TTM trial was not specifically designed for this sub-study, and the initial sampling of copeptin at 24 h may have been too late for optimal assessment of associations with outcome and circulatory derangement. Patients dying before 24 h were excluded from our analysis, limiting our results to patients alive at 24 h; also, the exclusion of patients in irreversible hemodynamic shock may limit the generalizability of our results.